Linear representations of grammaticality in neural language models

• URL: http://arxiv.org/abs/2607.15175v1
• Date: Thu, 16 Jul 2026 16:23:56 GMT
• Title: Linear representations of grammaticality in neural language models
• Abstract summary: We investigate whether grammaticality is encoded in the internal representations of neural language models.<n>Our results provide evidence that grammaticality is robustly encoded in sentence representations of a wide range of pretrained NLMs.
• Score: 7.771160777932259
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Whether neural language models (NLMs) possess the ability to distinguish strings on the basis of their grammaticality remains a debated topic in the computational linguistics literature. Existing evidence has largely relied on probability-based measures, testing whether models assign higher probabilities to grammatical than ungrammatical strings. However, probability comparisons have been criticized as a measure for grammatical knowledge based on the assumption that grammaticality is inherently entangled with likelihood. Model-assigned probability is a function of many related sentence properties, such as lexical frequency, plausibility, and world knowledge. In this work, we move beyond probability-based evaluations and investigate whether grammaticality is encoded in the internal representations of NLMs. Using mass-mean probing, we test whether grammatical and ungrammatical sentences are systematically separated in representational space. We further examine the extent to which these representations are independent of sentence properties that are correlated with grammaticality, as well as their generalization across grammatical phenomena and languages. Our results provide evidence that grammaticality is robustly encoded in sentence representations of a wide range of pretrained NLMs, yielding clear representational separation on the dimension of grammaticality that cannot be fully explained by alternative sentence-level factors. Moreover, this encoding generalizes across a broad range of grammatical phenomena and to some degree, across languages, suggesting that grammaticality constitutes a coherent representational dimension in contemporary NLMs. These findings contribute new evidence to debates about the nature of syntactic knowledge in language models and offer a complementary framework for evaluating grammatical competence that is not dependent on string probabilities alone.

Related papers

• Implicit Representations of Grammaticality in Language Models [33.563784182750474]
  We train a linear probe on a dataset of grammatical and (synthetic) ungrammatical sentences.<n>We find that this simple grammaticality probe generalizes to human-curated grammaticality judgment benchmarks and outperforms LM probability-based grammaticality judgments.<n>The English-trained probe also exhibits nontrivial cross-lingual generalization, outperforming string probabilities on grammaticality benchmarks in numerous other languages.
  arXiv  Detail & Related papers  (2026-05-06T17:57:31Z)
• Large Language Model probabilities cannot distinguish between possible and impossible language [0.11726720776908521]
  We use model-internal representations to tap directly into the way Large Language Models represent the 'grammatical-ungrammatical' distinction.<n>We predict that if string-probabilities can function as proxies for the limits of grammar, the ungrammatical condition will stand out among the conditions that involve linguistic violations.<n>Our results do not reveal a unique surprisal signature for ungrammatical prompts, as the semantically and pragmatically odd conditions consistently show higher surprisal.
  arXiv  Detail & Related papers  (2025-09-18T16:17:48Z)
• Principles of semantic and functional efficiency in grammatical patterning [1.6267479602370545]
  We show that grammatical organization provably inherits from perceptual attributes.<n>Our measurements on a diverse language sample show that grammars prioritize functional goals.
  arXiv  Detail & Related papers  (2024-10-21T10:49:54Z)
• Large Language Models can Contrastively Refine their Generation for Better Sentence Representation Learning [57.74233319453229]
  Large language models (LLMs) have emerged as a groundbreaking technology and their unparalleled text generation capabilities have sparked interest in their application to the fundamental sentence representation learning task. We propose MultiCSR, a multi-level contrastive sentence representation learning framework that decomposes the process of prompting LLMs to generate a corpus. Our experiments reveal that MultiCSR enables a less advanced LLM to surpass the performance of ChatGPT, while applying it to ChatGPT achieves better state-of-the-art results.
  arXiv  Detail & Related papers  (2023-10-17T03:21:43Z)
• CLSE: Corpus of Linguistically Significant Entities [58.29901964387952]
  We release a Corpus of Linguistically Significant Entities (CLSE) annotated by experts. CLSE covers 74 different semantic types to support various applications from airline ticketing to video games. We create a linguistically representative NLG evaluation benchmark in three languages: French, Marathi, and Russian.
  arXiv  Detail & Related papers  (2022-11-04T12:56:12Z)
• Transparency Helps Reveal When Language Models Learn Meaning [71.96920839263457]
  Our systematic experiments with synthetic data reveal that, with languages where all expressions have context-independent denotations, both autoregressive and masked language models learn to emulate semantic relations between expressions. Turning to natural language, our experiments with a specific phenomenon -- referential opacity -- add to the growing body of evidence that current language models do not well-represent natural language semantics.
  arXiv  Detail & Related papers  (2022-10-14T02:35:19Z)
• Testing the Ability of Language Models to Interpret Figurative Language [69.59943454934799]
  Figurative and metaphorical language are commonplace in discourse. It remains an open question to what extent modern language models can interpret nonliteral phrases. We introduce Fig-QA, a Winograd-style nonliteral language understanding task.
  arXiv  Detail & Related papers  (2022-04-26T23:42:22Z)
• A Latent-Variable Model for Intrinsic Probing [93.62808331764072]
  We propose a novel latent-variable formulation for constructing intrinsic probes.<n>We find empirical evidence that pre-trained representations develop a cross-lingually entangled notion of morphosyntax.
  arXiv  Detail & Related papers  (2022-01-20T15:01:12Z)
• Dependency Induction Through the Lens of Visual Perception [81.91502968815746]
  We propose an unsupervised grammar induction model that leverages word concreteness and a structural vision-based to jointly learn constituency-structure and dependency-structure grammars. Our experiments show that the proposed extension outperforms the current state-of-the-art visually grounded models in constituency parsing even with a smaller grammar size.
  arXiv  Detail & Related papers  (2021-09-20T18:40:37Z)
• Word Frequency Does Not Predict Grammatical Knowledge in Language Models [2.1984302611206537]
  We investigate whether there are systematic sources of variation in the language models' accuracy. We find that certain nouns are systematically understood better than others, an effect which is robust across grammatical tasks and different language models. We find that a novel noun's grammatical properties can be few-shot learned from various types of training data.
  arXiv  Detail & Related papers  (2020-10-26T19:51:36Z)
• Intrinsic Probing through Dimension Selection [69.52439198455438]
  Most modern NLP systems make use of pre-trained contextual representations that attain astonishingly high performance on a variety of tasks. Such high performance should not be possible unless some form of linguistic structure inheres in these representations, and a wealth of research has sprung up on probing for it. In this paper, we draw a distinction between intrinsic probing, which examines how linguistic information is structured within a representation, and the extrinsic probing popular in prior work, which only argues for the presence of such information by showing that it can be successfully extracted.
  arXiv  Detail & Related papers  (2020-10-06T15:21:08Z)
• Recurrent Neural Network Language Models Always Learn English-Like Relative Clause Attachment [17.995905582226463]
  We compare model performance in English and Spanish to show that non-linguistic biases in RNN LMs advantageously overlap with syntactic structure in English but not Spanish. English models may appear to acquire human-like syntactic preferences, while models trained on Spanish fail to acquire comparable human-like preferences.
  arXiv  Detail & Related papers  (2020-05-01T01:21:47Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.